HEAT TREATMENT. 383 by annealing if it is in contact with flame, while it is improved if it is reheated in a sealed muffle. I cannot assent to this broad conclusion, for, while it may be true that a flame can be run too hot and the piece be burned through carelessness, it by no means follows that such local overheating is necessary; nor is there any ground for assuming the absorption of deleterious gases from a proper flame. Moreover, the figures which he gives do not show a decided improvement of any kind in the bars which were heated in a retort. TABLE XV-L Comparative Physical Properties of Natural and Annealed Flat Steel Bars; as given by Henning.* w 05 is" t» on 2 . ef d 03 IM 8 2 3 +3'.,. A * d H <j) O J^S » v> o $2* .a ' '8w IM O S-s I. eS-u I Is verage of flats inches. o 3 1 TI'S 2 .2d S III 2-og -SflS :s§ sas1 S 03 d 1|8 gas -5CO Pi S » 0 « SM T^ 03 MS So ** Li ll fe B 4 o H P ps 8 . 10 I tolA 1.12 Natural Annealed 88787 40299 71220 69296 23.89 25.53 47.0 58.5 54.4 58.2 10 11 to 1& 1.41 Natural Annealed 85411 38298 68465 67971 24.88 24.95 46.65 49.17 51.7 56.3 12 11 to If 1.62 Natural Annealed 85729 88602 69490 69411 24.25 25.28 47.27 49.85 51.4 55.7 It is stated (loc. cit., p. 577) that most of the "flats" were "properly" annealed, and so I have averaged the records which he gives of the natural and the reheated pieces, separating them into three groups according to thickness. The results are given in Table XV-I. It will be seen that the metal has undergone very little change at all, and it is impossible to see anything which can be called a radical improvement. Any attempt to carry out a general system of annealing plates and shapes will result in wide variations in temperatures and rates of cooling, for it will be impossible to have a large pile of metal heated uniformly throughout, since the outside of the lot will be at * Trans. Amer. Soc. Mecli. Eng., Vol. XIII, p. 586, et seq. The factor which Mr. Henning calls the "yield point" Is here called the elastic limit I haTfc omitted from the averages the tests which are noted In the original as being wrongly marked, and also three tests which show such extremely low elongation that it Is certain the material waa not properly treated, or that there la an error in the records.